assert: deepEqual inconsistencies

[BridgeAR]

• Version: 8.1.0 - 8.x.x
• Subsystem: assert

Seems like #13318 introduced a regression where deep equal objects are now failing.

Two small test cases:

// First
// (This worked with 8.0.0 and before)
const a = {}
const b = {}
a.a = a
b.a = {}
b.a.a = a
assert.deepEqual(b, a) // ok
assert.deepEqual(a, b) // fails

// Second
// (This lead to a RangeError before the mentioned PR)
const a = new Set()
const b = new Set()
const c = new Set()
a.add(a)
b.add(b)
c.add(a)
assert.deepEqual(c, b) // ok
assert.deepEqual(b, c) // fails

I am not sure if I find the time to look into it in the next few days, so I opened the issue about it.

[XadillaX]

What's the expected result?

[Trott]

What's the expected result?

@XadillaX In the first case, the expected result is probably that no AssertionError is thrown (which is the behavior in 8.0.0), and certainly that the result of assert.deepEqual(a, b) is the same as assert.deepEqual(b, a).

In the second case, I think the behavior of both 8.0.0 ("Maximum call stack size exceeded") and current master branch (assert.deepEqual(c, b) is fine but assert.deepEqual(b, c) throws) are undesirable. It seems to me like no error should be thrown at all.

[Trott]

Proposed known-issue test for this in #14488